W3C home > Mailing lists > Public > www-international@w3.org > October to December 2002

Re: [IRI] IRI should not allow space; remove BNF from draft-duerst-iri-02...

From: Martin Duerst <duerst@w3.org>
Date: Wed, 27 Nov 2002 04:05:52 +0900
Message-Id: <4.2.0.58.J.20021127034636.0408fae8@localhost>
To: "Larry Masinter" <LMM@acm.org>, <www-international@w3.org>
Cc: <uri@w3.org>, Michel Suignard <michelsu@microsoft.com>

Hello Larry,

Many thanks for your comment on the IRI draft.

At 19:12 02/11/15 -0800, Larry Masinter wrote:

>I've heard the discussion about the inclusion of spaces
>in IRIs when they are not allowed in URIs, and I'm
>swayed by the argument that IRIs should only address
>the internationalization issue. That is, URIs and IRIs
>should not differ on where and when spaces are allowed.
>The only difference should be in non-ASCII characters.
>This makes the relationship between IRIs and URIs simpler.

I have heard similar arguments from other people.
The main reason to include spaces was because the
people working on XPointer wanted it. I'll make sure
they know about the new proposal and get a chance to
argue their way, on www-international@w3.org.

Another specific issue with IRIs is that there are many
kinds of spaces in non-ASCII (starting with NBSP). It
may be better to allow U+0020, leading to some degree
of transcribability, rather than forbid it, having
some people use other kinds of spaces, which are much
less transcribable.


>Perhaps there is some update in the URI draft
>(draft-fielding-uri-rfc2396bis-00.txt) that should
>talk about the use of interior spaces, but I don't
>thinkt he two documents should differ.

Using actual spaces (rather than %20) seems to be
rather popular in fragment identifiers.


>Secondly, note that the RFC 2396 update will
>also update the BNF for URIs, and the IRI draft
>should follow.

As you may have noted,
http://www.ietf.org/internet-drafts/draft-duerst-iri-02.txt
already updates the ABNF in parallel with rfc2396bis.
We have done this based on a request from Roy.


>I think that it would be a shame
>to issue the IRI RFC and then soon after issue
>a revised URI RFC with a different BNF.

Yes, that would probably be a bad idea, in particular
if there are big differences.


>So I think the solution to this is to remove
>the BNF from section 2.2 of the IRI draft.
>(It was a good idea at the time, but the
>timing is now wrong).

Why do you think it's a bad idea? Because there is a
risk that there will be some more updates to rfc2396bis
while the IRI draft will already have moved to proposed?

How do you suggest we express the various details and
restrictions on IRI syntax without an ABNF?
As an example, we do not allow private use codepoints
in most parts of an IRI, but we allow them in the query
part.


Regards,    Martin.
Received on Tuesday, 26 November 2002 14:06:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 19:16:59 GMT